Yeah, that was one of my first thoughts too. With Time Machine currently offering no way to control the frequency of back-ups, my 2009 Core 2 Duo iMac gets virtually paralyzed once an hour for what seems like 10 minutes of back up. And that is with a FireWire 800 connector! When Western Digital comes out with a Thunderbolt MyBook I'm buying both it and a new iMac. On the other hand, if Lion is tied to iCloud, there may be other alternatives in play later this year.
you can download a piece of software called Time Machine Editor that allows you to change the frequency of your backups.
While I'm not a huge fan of benchmarks, a machine that is all around twice as fast will be twice as fast for all users.
That's absolutely not true.
"Twice as fast" requires details. For example, let's say that it was twice as fast at Geekbench. That means that any applications which use the computer similarly to Geekbench will also be about twice as fast. But if the application uses the computer differently, it could be more or less than twice as fast.
For example:
Let's say that the new computer is twice as fast in Geekbench. Geekbench uses the CPU and memory, but not the hard disk, FPU, or graphics system.
Now, let's say that when you measure the new computer's components separately, the new computer is:
50% faster on CPU and 150% faster on memory. Geekbench would say it's 100% faster, on average (assuming equal weighting) - or twice as fast.
But let's say it's 2000% faster on graphics, 20% faster on hard disk, and 4000% faster on FPU. If the user is doing something that stresses the entire computer, they will perceive much more than twice as fast. Similarly, if the hard disk is only 20% faster (which might be reasonable) and they're doing something that is hard disk limited, then the new system will not be anywhere close to twice as fast - it might be closer to 20% faster. Thus, some users might see much more than twice as fast and others might see much less.
Benchmarks are meaningless unless you understand what you're measuring - and they are only useful if the applications you use have similar properties to the benchmark.
You missed the point . unless someone does the same stressful action over and over , over time say a year >>on average it will be twice as fast . not every time but over time .
That is absolutely, 100% false.
Look at the above examples: Geekbench measures CPU and memory performance. If that's all you do, then Geekbench MAY BE a reasonable benchmark for your work (you'd still have to verity that it's testing the same functions that you use and that the weighting reflects your work).
But let's say you are into heavy gaming. Game performance depends more on the video card than anything else, particularly at high resolution. The new system being 100% faster on CPU and memory access tells you nothing about graphics performance. The new system might be 300% faster ... or 1,000 percent faster... or 2% faster than the old one. Geekbench will tell you absolutely NOTHING about the system's performance on games.
Or let's say your work involves a lot of file transfers. Say the CPU part is done in milliseconds and then it takes seconds to transfer the file to the hard disk. Geekbench would not be useful. If the new hard disk is no faster in the new system, then the user who is hard disk limited would not see any benefit - even though Geekbench says it's twice as fast. OTOH, if the new disk has an SSD, the user who is hard disk limited is going to see a lot more than 100% speed gains.
Benchmarks measure only a limited set of performance criteria. Even if the benchmark does hit the CPU, FPU, memory, hard disk, optical disk, and graphics, it STILL may not be useful for evaluating any given user's needs because every users's needs differ - and even a benchmark claiming to test all of the above functions only tests a subset of them. And then you have to worry about weighting.
You couldn't be any more wrong. Perhaps you should stop posting until you learn about the topics you want to post about.
Look at the above examples: Geekbench measures CPU and memory performance. If that's all you do, then Geekbench MAY BE a reasonable benchmark for your work (you'd still have to verity that it's testing the same functions that you use and that the weighting reflects your work).
But let's say you are into heavy gaming. Game performance depends more on the video card than anything else, particularly at high resolution. The new system being 100% faster on CPU and memory access tells you nothing about graphics performance. The new system might be 300% faster ... or 1,000 percent faster... or 2% faster than the old one. Geekbench will tell you absolutely NOTHING about the system's performance on games.
Or let's say your work involves a lot of file transfers. Say the CPU part is done in milliseconds and then it takes seconds to transfer the file to the hard disk. Geekbench would not be useful. If the new hard disk is no faster in the new system, then the user who is hard disk limited would not see any benefit - even though Geekbench says it's twice as fast. OTOH, if the new disk has an SSD, the user who is hard disk limited is going to see a lot more than 100% speed gains.
Benchmarks measure only a limited set of performance criteria. Even if the benchmark does hit the CPU, FPU, memory, hard disk, optical disk, and graphics, it STILL may not be useful for evaluating any given user's needs because every users's needs differ - and even a benchmark claiming to test all of the above functions only tests a subset of them. And then you have to worry about weighting.
You couldn't be any more wrong. Perhaps you should stop posting until you learn about the topics you want to post about.
hey cool down dude not that big a deal either way . the geeks say 2x as fast . if the geek test is off not my fault . maybe i am wrong . i will re read the whole thing .
Look at the above examples: Geekbench measures CPU and memory performance. If that's all you do, then Geekbench MAY BE a reasonable benchmark for your work (you'd still have to verity that it's testing the same functions that you use and that the weighting reflects your work).
But let's say you are into heavy gaming. Game performance depends more on the video card than anything else, particularly at high resolution. The new system being 100% faster on CPU and memory access tells you nothing about graphics performance. The new system might be 300% faster ... or 1,000 percent faster... or 2% faster than the old one. Geekbench will tell you absolutely NOTHING about the system's performance on games.
Or let's say your work involves a lot of file transfers. Say the CPU part is done in milliseconds and then it takes seconds to transfer the file to the hard disk. Geekbench would not be useful. If the new hard disk is no faster in the new system, then the user who is hard disk limited would not see any benefit - even though Geekbench says it's twice as fast. OTOH, if the new disk has an SSD, the user who is hard disk limited is going to see a lot more than 100% speed gains.
Benchmarks measure only a limited set of performance criteria. Even if the benchmark does hit the CPU, FPU, memory, hard disk, optical disk, and graphics, it STILL may not be useful for evaluating any given user's needs because every users's needs differ - and even a benchmark claiming to test all of the above functions only tests a subset of them. And then you have to worry about weighting.
You couldn't be any more wrong. Perhaps you should stop posting until you learn about the topics you want to post about.
QFT.
You can't simply say 1 computer is X% faster than another because different tasks or programs stress different aspects of a computer. Some might be GPU bound, some may be CPU bound etc.
The best way to 'benchmark' one computer against another is to break each down to its individual components, and test the difference in performance of each component in a controlled environment (ie. keeping all other components equal and making sure no other component is bottlenecking the test).
hey cool down dude not that big a deal either way . the geeks say 2x as fast . if the geek test is off not my fault . maybe i am wrong . i will re read the whole thing .
It's not that the geek test is off, its that you're drawing conclusions thats not supported by the test. I have no doubt that they tested 2x as fast using geekbench, but that's all you can say about it.
Look at it this way, say you have 2 cars, one with 200 horsepower, and 1 with 400 horsepower. From that alone, you cannot say the car with 400 horsepower will be twice as fast in all situations. You have to take into account weight, gearing, etc. Even then, you have to specify what kind of race it is. Is it a straight line drag race? a road course that stresses suspension setup and braking?
It's the same thing for the computer -- the CPU may be 2x as fast (though honestly, I don't know what geekbench tests), but you still have to take into account all the other components like memory, GPU, hard disk, etc. Even then, you have to specify the task, as different tasks stress different components.
Even when talking about single components, its not that black and white. Here is a compilation of benchmarks between the old 2.93 Ghz CPU and the new 3.4 Ghz CPU: http://www.anandtech.com/bench/Product/287?vs=107
notice the wide range of performance differences. This occurs because each benchmark tests different aspects of the CPU. Some tests (such as productively) stress single thread performance, others (like video encoding) are hugely parallel and scale well with multiple cores, some tests are greatly affected by cache, etc etc.
hey cool down dude not that big a deal either way . the geeks say 2x as fast . if the geek test is off not my fault . maybe i am wrong . i will re read the whole thing .
Misleading. For some tasks for some people it could be up to twice as fast.
But for most tasks for most people there's not much of a difference.
Look at what they can do on an iPad with a much lesser cpu than a C2D and you'll see where I'm coming from.
For all my cynicism nowadays it's pretty amazing what Intel has achieved with Sandy Bridge and in general since the "dark days" of CPU manufacturers hitting the 90nm wall and the fiasco of the PowerPC G5 and Pentium 4.
The irony is that, as you say, most tasks nowadays for most people don't need the level of horsepower of the current Intel CPUs.
And we have seen ARM rise to the forefront and developers do some kickass stuff with that. Within 3 years at the rate Apple, ARM and the others are going, we could see ARM laptops do most of the work people need.
The only issue is that only Apple has really figured out the software side of the equation in getting the absolute maximum performance and USABILITY out of ARM in a coherent package.
For all my cynicism nowadays it's pretty amazing what Intel has achieved with Sandy Bridge and in general since the "dark days" of CPU manufacturers hitting the 90nm wall and the fiasco of the PowerPC G5 and Pentium 4.
The irony is that, as you say, most tasks nowadays for most people don't need the level of horsepower of the current Intel CPUs.
That has been true for years - but hasn't stopped "newer" and "faster" and "umpteen gigahertz" from being a selling point.
It's like cars. Most people could get by quite nicely with 100 HP (or less). Yet you see almost every car brand bragging about how much more horsepower they have than the competition - or how much faster they are than last year's model.
People are driven toward 'new and improved'. Droping to a much slower machine would be a hard sell unless there's a compelling advantage. Apple showed a compelling advantage with the iPad - and then with the MacBook Air, so people could live with the performance. But what would be the compelling advantage to cut performance significantly this time? Battery life? Nope - they're already at 10 hours, so battery life isn't much of an issue (especially since Ivy Bridge will increase that further).
Quote:
Originally Posted by nvidia2008
And we have seen ARM rise to the forefront and developers do some kickass stuff with that. Within 3 years at the rate Apple, ARM and the others are going, we could see ARM laptops do most of the work people need.
See above.
More importantly, you're assuming that ARM will increase performance significantly without increasing power consumption. Intel has a massive lead in performance right now with ARM having an advantage in power consumption. As ARM increases performance to approach Intel's low end, their power consumption will go up. Meanwhile, Intel has made some significant breakthroughs in power consumption. Ivy Bridge will be very impressive in that regard.
While ARM will likely be somewhat closer to Intel's performance in 3 years than they are today, the power consumption savings will likely be far less - probably not enough to justify the architecture change.
Note, too, that CPU power is probably the lowest percentage of the entire power budget that it's ever been. Even if the CPU drew zero power, you still have backlighting, hard disk or SSD, optical drive (if present), GPU (if not included in the CPU), LCD display, WiFi, and so on. As CPUs get more efficient, their portion of the total declines, which means that potential savings decline, as well.
Quote:
Originally Posted by nvidia2008
The only issue is that only Apple has really figured out the software side of the equation in getting the absolute maximum performance and USABILITY out of ARM in a coherent package.
That's not the real issue. Apple has already shown that they can take advantage of ARM quite well. The real issue is whether other software vendors are willing to do so.
The big question is, why is it that Apple can deliver a quad-core Sandy Bridge desktop with a 21.5 inch quality LCD display and input devices along with a decent GPU, for (using Canadian prices) $1,199, yet the Mini checks in at $699 yet has dramatically inferior internals.
If you do as much upgrading as Apple will allow to try and get the Mini as close to the iMac as possible, something amazing happens. The Mini with an upgrade to a 2,66Ghz Core 2 Duo processor, a 500Gb 5,400 RPM drive, 4GB of RAM, a magic trackpad and a wired keyboard, retails for $1,132.
The base iMac comes with a quad-core Sandy Bridge clocked at 2.5Ghz, 4GB of RAM, a 7,200 RPM hard drive.
It's outrageous that Apple would charge $68 less for far less capable hardware and then add a high-quality monitor and a 17-watt stereo system virtually for free.
The fundamental problem is that laptop components cost more than the pieces used in the iMac. After all, the Macbook Pro doesn't compare well to the iMac, either. It has less power, a smaller, slower hard drive, and a smaller screen yet costs more. Some of that is due to battery technology not required in a desktop, the Mini included.
But let's suppose we accept the notion that the Mini, using as it does costlier laptop hardware, is bound to cost more than a true desktop like the iMac. Using that approach, let's consider what the machine would look like cost-wise if it amounted to a desktop version of the 15" Macbook Pro. That machine in base form is priced in Canada at $1,849. That's for 4GB of RAM and two GPUs, namely the integrated Intel graphics and an additional HD6490 Radeon.
To take similar tech and use it in a desktop, first of all, there would be no need for two GPUs because power considerations would not matter in a device that is running off of household current all the time. No need for a battery, either. I would imagine this would result in a savings of let's say $300. No need for a screen which would take maybe $250 off. No input devices, so take another $100 off. No camera, lose another $50. Far less capable sound with a mono speaker running at low power vs. a stereo setup with more juice. There's about $75 coming off. All of those items gone would add up to a reduction of $775 which would bring the Mini in at $1,074. But for that price you'd get 4GB of RAM, a 2.0Ghz quad-core processor, and a 500GB drive. To get 4GB RAM and a 500 GB HD on the current Mini sets you back $1,014 if you also upgrade to a 2.66Ghz Core 2 Duo processor.
There are two questions that come to mind. The first is why Apple continues to use costlier laptop pieces in the Mini when there are more cost-effective options like those worked into the iMac which pretty much has it's components built into the body housing the monitor. The other is, why is Apple currently charging so much for a device using older technology.
The problem I have with Apple's current approach is that having a fairly new Cinema Display, I'm just not interested in a all-in-one design like the iMac yet I can't justify spending close to $3,000 on a Mac Pro considering I'm using the equipment just to pursue a few hobbies. So for me, I'd rather that there was what amounted to a headless iMac rather than pay a laptop premium for a unit that is basically a laptop minus the portability and of course a dedicated screen. I'd rather have iMac performance in a device that doesn't come with a monitor.
By the way, perhaps I'm mistaken but doesn't Thunderbolt mean that a device similar to the Mini but more closely related to the iMac could be expanded because of a fast interface that would allow stuff like running the OS on a faster external drive, dramatically expanding memory with no speed penalty, etc.
It's not that the geek test is off, its that you're drawing conclusions thats not supported by the test. I have no doubt that they tested 2x as fast using geekbench, but that's all you can say about it.
Look at it this way, say you have 2 cars, one with 200 horsepower, and 1 with 400 horsepower. From that alone, you cannot say the car with 400 horsepower will be twice as fast in all situations. You have to take into account weight, gearing, etc. Even then, you have to specify what kind of race it is. Is it a straight line drag race? a road course that stresses suspension setup and braking?
It's the same thing for the computer -- the CPU may be 2x as fast (though honestly, I don't know what geekbench tests), but you still have to take into account all the other components like memory, GPU, hard disk, etc. Even then, you have to specify the task, as different tasks stress different components.
Even when talking about single components, its not that black and white. Here is a compilation of benchmarks between the old 2.93 Ghz CPU and the new 3.4 Ghz CPU: http://www.anandtech.com/bench/Product/287?vs=107
notice the wide range of performance differences. This occurs because each benchmark tests different aspects of the CPU. Some tests (such as productively) stress single thread performance, others (like video encoding) are hugely parallel and scale well with multiple cores, some tests are greatly affected by cache, etc etc.
I read all the posts .I ALL the post every day about chips and the such .
I simply must have missed the meaning or you are smarter than me about chips . Either way i do no care for your tone at all .
That's OK. I don't care for people who insist on making public displays of their own ignorance - and then when they're corrected, attacking the person who tries to educate them. I also "do no care" for people who write like a 2nd grader.
Comments
Yeah, that was one of my first thoughts too. With Time Machine currently offering no way to control the frequency of back-ups, my 2009 Core 2 Duo iMac gets virtually paralyzed once an hour for what seems like 10 minutes of back up. And that is with a FireWire 800 connector! When Western Digital comes out with a Thunderbolt MyBook I'm buying both it and a new iMac. On the other hand, if Lion is tied to iCloud, there may be other alternatives in play later this year.
you can download a piece of software called Time Machine Editor that allows you to change the frequency of your backups.
So what would a 2.8 P4 compare to one of these? Man I need to upgrade.
Well, I couldn't find a 2.8 P4, but here's now a 3.6 P4 compares to an i7 2600k:
http://www.anandtech.com/bench/Product/92?vs=287
While I'm not a huge fan of benchmarks, a machine that is all around twice as fast will be twice as fast for all users.
That's absolutely not true.
"Twice as fast" requires details. For example, let's say that it was twice as fast at Geekbench. That means that any applications which use the computer similarly to Geekbench will also be about twice as fast. But if the application uses the computer differently, it could be more or less than twice as fast.
For example:
Let's say that the new computer is twice as fast in Geekbench. Geekbench uses the CPU and memory, but not the hard disk, FPU, or graphics system.
Now, let's say that when you measure the new computer's components separately, the new computer is:
50% faster on CPU and 150% faster on memory. Geekbench would say it's 100% faster, on average (assuming equal weighting) - or twice as fast.
But let's say it's 2000% faster on graphics, 20% faster on hard disk, and 4000% faster on FPU. If the user is doing something that stresses the entire computer, they will perceive much more than twice as fast. Similarly, if the hard disk is only 20% faster (which might be reasonable) and they're doing something that is hard disk limited, then the new system will not be anywhere close to twice as fast - it might be closer to 20% faster. Thus, some users might see much more than twice as fast and others might see much less.
Benchmarks are meaningless unless you understand what you're measuring - and they are only useful if the applications you use have similar properties to the benchmark.
You missed the point . unless someone does the same stressful action over and over , over time say a year >>on average it will be twice as fast . not every time but over time .
That is absolutely, 100% false.
Look at the above examples: Geekbench measures CPU and memory performance. If that's all you do, then Geekbench MAY BE a reasonable benchmark for your work (you'd still have to verity that it's testing the same functions that you use and that the weighting reflects your work).
But let's say you are into heavy gaming. Game performance depends more on the video card than anything else, particularly at high resolution. The new system being 100% faster on CPU and memory access tells you nothing about graphics performance. The new system might be 300% faster ... or 1,000 percent faster... or 2% faster than the old one. Geekbench will tell you absolutely NOTHING about the system's performance on games.
Or let's say your work involves a lot of file transfers. Say the CPU part is done in milliseconds and then it takes seconds to transfer the file to the hard disk. Geekbench would not be useful. If the new hard disk is no faster in the new system, then the user who is hard disk limited would not see any benefit - even though Geekbench says it's twice as fast. OTOH, if the new disk has an SSD, the user who is hard disk limited is going to see a lot more than 100% speed gains.
Benchmarks measure only a limited set of performance criteria. Even if the benchmark does hit the CPU, FPU, memory, hard disk, optical disk, and graphics, it STILL may not be useful for evaluating any given user's needs because every users's needs differ - and even a benchmark claiming to test all of the above functions only tests a subset of them. And then you have to worry about weighting.
You couldn't be any more wrong. Perhaps you should stop posting until you learn about the topics you want to post about.
That is absolutely, 100% false.
Look at the above examples: Geekbench measures CPU and memory performance. If that's all you do, then Geekbench MAY BE a reasonable benchmark for your work (you'd still have to verity that it's testing the same functions that you use and that the weighting reflects your work).
But let's say you are into heavy gaming. Game performance depends more on the video card than anything else, particularly at high resolution. The new system being 100% faster on CPU and memory access tells you nothing about graphics performance. The new system might be 300% faster ... or 1,000 percent faster... or 2% faster than the old one. Geekbench will tell you absolutely NOTHING about the system's performance on games.
Or let's say your work involves a lot of file transfers. Say the CPU part is done in milliseconds and then it takes seconds to transfer the file to the hard disk. Geekbench would not be useful. If the new hard disk is no faster in the new system, then the user who is hard disk limited would not see any benefit - even though Geekbench says it's twice as fast. OTOH, if the new disk has an SSD, the user who is hard disk limited is going to see a lot more than 100% speed gains.
Benchmarks measure only a limited set of performance criteria. Even if the benchmark does hit the CPU, FPU, memory, hard disk, optical disk, and graphics, it STILL may not be useful for evaluating any given user's needs because every users's needs differ - and even a benchmark claiming to test all of the above functions only tests a subset of them. And then you have to worry about weighting.
You couldn't be any more wrong. Perhaps you should stop posting until you learn about the topics you want to post about.
hey cool down dude not that big a deal either way . the geeks say 2x as fast . if the geek test is off not my fault . maybe i am wrong . i will re read the whole thing .
peace
9
That is absolutely, 100% false.
Look at the above examples: Geekbench measures CPU and memory performance. If that's all you do, then Geekbench MAY BE a reasonable benchmark for your work (you'd still have to verity that it's testing the same functions that you use and that the weighting reflects your work).
But let's say you are into heavy gaming. Game performance depends more on the video card than anything else, particularly at high resolution. The new system being 100% faster on CPU and memory access tells you nothing about graphics performance. The new system might be 300% faster ... or 1,000 percent faster... or 2% faster than the old one. Geekbench will tell you absolutely NOTHING about the system's performance on games.
Or let's say your work involves a lot of file transfers. Say the CPU part is done in milliseconds and then it takes seconds to transfer the file to the hard disk. Geekbench would not be useful. If the new hard disk is no faster in the new system, then the user who is hard disk limited would not see any benefit - even though Geekbench says it's twice as fast. OTOH, if the new disk has an SSD, the user who is hard disk limited is going to see a lot more than 100% speed gains.
Benchmarks measure only a limited set of performance criteria. Even if the benchmark does hit the CPU, FPU, memory, hard disk, optical disk, and graphics, it STILL may not be useful for evaluating any given user's needs because every users's needs differ - and even a benchmark claiming to test all of the above functions only tests a subset of them. And then you have to worry about weighting.
You couldn't be any more wrong. Perhaps you should stop posting until you learn about the topics you want to post about.
QFT.
You can't simply say 1 computer is X% faster than another because different tasks or programs stress different aspects of a computer. Some might be GPU bound, some may be CPU bound etc.
The best way to 'benchmark' one computer against another is to break each down to its individual components, and test the difference in performance of each component in a controlled environment (ie. keeping all other components equal and making sure no other component is bottlenecking the test).
hey cool down dude not that big a deal either way . the geeks say 2x as fast . if the geek test is off not my fault . maybe i am wrong . i will re read the whole thing .
It's not that the geek test is off, its that you're drawing conclusions thats not supported by the test. I have no doubt that they tested 2x as fast using geekbench, but that's all you can say about it.
Look at it this way, say you have 2 cars, one with 200 horsepower, and 1 with 400 horsepower. From that alone, you cannot say the car with 400 horsepower will be twice as fast in all situations. You have to take into account weight, gearing, etc. Even then, you have to specify what kind of race it is. Is it a straight line drag race? a road course that stresses suspension setup and braking?
It's the same thing for the computer -- the CPU may be 2x as fast (though honestly, I don't know what geekbench tests), but you still have to take into account all the other components like memory, GPU, hard disk, etc. Even then, you have to specify the task, as different tasks stress different components.
Even when talking about single components, its not that black and white. Here is a compilation of benchmarks between the old 2.93 Ghz CPU and the new 3.4 Ghz CPU: http://www.anandtech.com/bench/Product/287?vs=107
notice the wide range of performance differences. This occurs because each benchmark tests different aspects of the CPU. Some tests (such as productively) stress single thread performance, others (like video encoding) are hugely parallel and scale well with multiple cores, some tests are greatly affected by cache, etc etc.
hey cool down dude not that big a deal either way . the geeks say 2x as fast . if the geek test is off not my fault . maybe i am wrong . i will re read the whole thing .
Once again, learn first, and THEN post.
Misleading. For some tasks for some people it could be up to twice as fast.
But for most tasks for most people there's not much of a difference.
Look at what they can do on an iPad with a much lesser cpu than a C2D and you'll see where I'm coming from.
For all my cynicism nowadays it's pretty amazing what Intel has achieved with Sandy Bridge and in general since the "dark days" of CPU manufacturers hitting the 90nm wall and the fiasco of the PowerPC G5 and Pentium 4.
The irony is that, as you say, most tasks nowadays for most people don't need the level of horsepower of the current Intel CPUs.
And we have seen ARM rise to the forefront and developers do some kickass stuff with that. Within 3 years at the rate Apple, ARM and the others are going, we could see ARM laptops do most of the work people need.
The only issue is that only Apple has really figured out the software side of the equation in getting the absolute maximum performance and USABILITY out of ARM in a coherent package.
Interesting times.
For all my cynicism nowadays it's pretty amazing what Intel has achieved with Sandy Bridge and in general since the "dark days" of CPU manufacturers hitting the 90nm wall and the fiasco of the PowerPC G5 and Pentium 4.
The irony is that, as you say, most tasks nowadays for most people don't need the level of horsepower of the current Intel CPUs.
That has been true for years - but hasn't stopped "newer" and "faster" and "umpteen gigahertz" from being a selling point.
It's like cars. Most people could get by quite nicely with 100 HP (or less). Yet you see almost every car brand bragging about how much more horsepower they have than the competition - or how much faster they are than last year's model.
People are driven toward 'new and improved'. Droping to a much slower machine would be a hard sell unless there's a compelling advantage. Apple showed a compelling advantage with the iPad - and then with the MacBook Air, so people could live with the performance. But what would be the compelling advantage to cut performance significantly this time? Battery life? Nope - they're already at 10 hours, so battery life isn't much of an issue (especially since Ivy Bridge will increase that further).
And we have seen ARM rise to the forefront and developers do some kickass stuff with that. Within 3 years at the rate Apple, ARM and the others are going, we could see ARM laptops do most of the work people need.
See above.
More importantly, you're assuming that ARM will increase performance significantly without increasing power consumption. Intel has a massive lead in performance right now with ARM having an advantage in power consumption. As ARM increases performance to approach Intel's low end, their power consumption will go up. Meanwhile, Intel has made some significant breakthroughs in power consumption. Ivy Bridge will be very impressive in that regard.
While ARM will likely be somewhat closer to Intel's performance in 3 years than they are today, the power consumption savings will likely be far less - probably not enough to justify the architecture change.
Note, too, that CPU power is probably the lowest percentage of the entire power budget that it's ever been. Even if the CPU drew zero power, you still have backlighting, hard disk or SSD, optical drive (if present), GPU (if not included in the CPU), LCD display, WiFi, and so on. As CPUs get more efficient, their portion of the total declines, which means that potential savings decline, as well.
The only issue is that only Apple has really figured out the software side of the equation in getting the absolute maximum performance and USABILITY out of ARM in a coherent package.
That's not the real issue. Apple has already shown that they can take advantage of ARM quite well. The real issue is whether other software vendors are willing to do so.
If you do as much upgrading as Apple will allow to try and get the Mini as close to the iMac as possible, something amazing happens. The Mini with an upgrade to a 2,66Ghz Core 2 Duo processor, a 500Gb 5,400 RPM drive, 4GB of RAM, a magic trackpad and a wired keyboard, retails for $1,132.
The base iMac comes with a quad-core Sandy Bridge clocked at 2.5Ghz, 4GB of RAM, a 7,200 RPM hard drive.
It's outrageous that Apple would charge $68 less for far less capable hardware and then add a high-quality monitor and a 17-watt stereo system virtually for free.
The fundamental problem is that laptop components cost more than the pieces used in the iMac. After all, the Macbook Pro doesn't compare well to the iMac, either. It has less power, a smaller, slower hard drive, and a smaller screen yet costs more. Some of that is due to battery technology not required in a desktop, the Mini included.
But let's suppose we accept the notion that the Mini, using as it does costlier laptop hardware, is bound to cost more than a true desktop like the iMac. Using that approach, let's consider what the machine would look like cost-wise if it amounted to a desktop version of the 15" Macbook Pro. That machine in base form is priced in Canada at $1,849. That's for 4GB of RAM and two GPUs, namely the integrated Intel graphics and an additional HD6490 Radeon.
To take similar tech and use it in a desktop, first of all, there would be no need for two GPUs because power considerations would not matter in a device that is running off of household current all the time. No need for a battery, either. I would imagine this would result in a savings of let's say $300. No need for a screen which would take maybe $250 off. No input devices, so take another $100 off. No camera, lose another $50. Far less capable sound with a mono speaker running at low power vs. a stereo setup with more juice. There's about $75 coming off. All of those items gone would add up to a reduction of $775 which would bring the Mini in at $1,074. But for that price you'd get 4GB of RAM, a 2.0Ghz quad-core processor, and a 500GB drive. To get 4GB RAM and a 500 GB HD on the current Mini sets you back $1,014 if you also upgrade to a 2.66Ghz Core 2 Duo processor.
There are two questions that come to mind. The first is why Apple continues to use costlier laptop pieces in the Mini when there are more cost-effective options like those worked into the iMac which pretty much has it's components built into the body housing the monitor. The other is, why is Apple currently charging so much for a device using older technology.
The problem I have with Apple's current approach is that having a fairly new Cinema Display, I'm just not interested in a all-in-one design like the iMac yet I can't justify spending close to $3,000 on a Mac Pro considering I'm using the equipment just to pursue a few hobbies. So for me, I'd rather that there was what amounted to a headless iMac rather than pay a laptop premium for a unit that is basically a laptop minus the portability and of course a dedicated screen. I'd rather have iMac performance in a device that doesn't come with a monitor.
By the way, perhaps I'm mistaken but doesn't Thunderbolt mean that a device similar to the Mini but more closely related to the iMac could be expanded because of a fast interface that would allow stuff like running the OS on a faster external drive, dramatically expanding memory with no speed penalty, etc.
It's not that the geek test is off, its that you're drawing conclusions thats not supported by the test. I have no doubt that they tested 2x as fast using geekbench, but that's all you can say about it.
Look at it this way, say you have 2 cars, one with 200 horsepower, and 1 with 400 horsepower. From that alone, you cannot say the car with 400 horsepower will be twice as fast in all situations. You have to take into account weight, gearing, etc. Even then, you have to specify what kind of race it is. Is it a straight line drag race? a road course that stresses suspension setup and braking?
It's the same thing for the computer -- the CPU may be 2x as fast (though honestly, I don't know what geekbench tests), but you still have to take into account all the other components like memory, GPU, hard disk, etc. Even then, you have to specify the task, as different tasks stress different components.
Even when talking about single components, its not that black and white. Here is a compilation of benchmarks between the old 2.93 Ghz CPU and the new 3.4 Ghz CPU: http://www.anandtech.com/bench/Product/287?vs=107
notice the wide range of performance differences. This occurs because each benchmark tests different aspects of the CPU. Some tests (such as productively) stress single thread performance, others (like video encoding) are hugely parallel and scale well with multiple cores, some tests are greatly affected by cache, etc etc.
Thanks dude .
9
be nice to me ok
I read all the posts .I ALL the post every day about chips and the such .
I simply must have missed the meaning or you are smarter than me about chips . Either way i do no care for your tone at all .
That's OK. I don't care for people who insist on making public displays of their own ignorance - and then when they're corrected, attacking the person who tries to educate them. I also "do no care" for people who write like a 2nd grader.
So what would a 2.8 P4 compare to one of these? Man I need to upgrade.
why ??